Dear All, I have a problem in using "mirclassify" function. I have several sounds episodes which have lengths of about 5 sec each. Extracting some features from those episodes' frames (50 msec) I obtain 100 by n (feature vector size) matrices. At this point, my problem is how to give the inputs to mirclassify function: "mirclassify(test, mfcc(test), train, mfcc(train))" *test: * The audio to be tested.. ( Does it have to be a single mirobject or can I use a series of mirobjects? ) * * * * *mfcc(test) :* For my problem I am not using mfcc but some extra features which are converted in to mirobject format. If I use a series of mir object ( taken from miraudio) do I have to use seperate feature vectors or does it have to be a single one which includes all the features? Best Regards Ersin Karcı